When an n8n workflow silently stops (or silently does nothing)

Three ways n8n fails without an error

• The trigger never fires. The instance restarted, the container ran out of memory, the queue worker died, or the workflow was deactivated during an edit. No execution means no Error Trigger.
• "Success" with nothing in it. An HTTP Request node returns 200 OK with an empty array, an API changes shape, or an IF node routes everything to the false branch. The execution is green; the output is empty.
• Self-hosted drift. Credentials expire, a webhook URL changes after a domain move, or the host's clock and timezone don't match what you think.

Add a heartbeat at the end of the workflow

Add an HTTP Request node as the final step that POSTs to your Did It Run? ping URL with the number of items processed:

{ "items": {{ $input.all().length }}, "text": "run complete" }

Create the watch with the same schedule as your Schedule Trigger and a grace window that covers normal run time. If the ping doesn't arrive by the deadline, you're alerted by email, Slack, Discord or Telegram.

Turn "green but empty" into an alert

Set Minimum item count to 1 (or whatever is normal). A run that processes zero items now raises an empty alert instead of a false sense of security. You can also require a keyword or a minimum body size.

Get failures instantly too

Point your error workflow (Error Trigger → HTTP Request) at /fail on the same ping URL. Failures alert right away; missed runs alert at the deadline.

Monitoring many workflows

The daily roll-call digest lists every workflow that ran, every one that's missing, and every one that came back empty — one email instead of opening the executions list for each workflow.